SEPURI-SAI-KRISHNA commented on PR #13092:
URL: https://github.com/apache/seatunnel/pull/13092#issuecomment-6897169194

   @DanielLeens @nzw921rx Thanks, both points are done in `6a59aa1a32`:
   
   1. The `partition_column` row (EN and ZH) no longer says NULL rows are never 
read. It now says that when the table is split by this column, a row with NULL 
in it falls in no split range or hash bucket and is not read; that with 
`enable_concurrent_read = false` the table is not split, so these rows are 
read; and that CockroachDB hash splits put NULL into one bucket 
(`CockroachDBDialect#hashModForField`). The split key section now states the 
contract: the automatic choice takes nullability from the table metadata, so it 
applies only to `table_path` reads (`findSplitKey` keeps a `query` without 
`partition_column` on one split), and an explicit `partition_column` is used as 
configured. The Javadoc of `isAutomaticSplitColumn` says the same.
   2. `JdbcMysqlSplitIT#readEveryIdOnce` has a Javadoc: the fixture (ids 1 to 
100, every tenth code NULL), splits planned from the real MySQL catalog 
metadata, every split query run, a duplicate id failing at once and a missing 
one failing the count of 100.
   
   The `incompatible-changes.md` entry now also names the primary key and the 
WARN. Docs and comments only; `DynamicChunkSplitterTest` 15/15 and 
`MarkdownTest` 4/4 pass, and the PR description is updated.
   
   @nzw921rx this push dismisses your approval; could you approve again if the 
wording is fine?
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to